thermal conductivity of silicon nitride membranes is not sensitive … · 2021. 1. 18. · the...
TRANSCRIPT
-
HAL Id: hal-01231839https://hal.archives-ouvertes.fr/hal-01231839
Submitted on 25 Nov 2015
HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.
Thermal conductivity of silicon nitride membranes is notsensitive to stress
Hossein Ftouni, Christophe Blanc, Dimitri Tainoff, Andrew Fefferman,Martial Defoort, Kunal Lulla, Jacques Richard, Eddy Collin, Olivier Bourgeois
To cite this version:Hossein Ftouni, Christophe Blanc, Dimitri Tainoff, Andrew Fefferman, Martial Defoort, et al.. Ther-mal conductivity of silicon nitride membranes is not sensitive to stress. Physical Review B: CondensedMatter and Materials Physics (1998-2015), American Physical Society, 2015, 92 (12), pp.125439.�10.1103/PhysRevB.92.125439�. �hal-01231839�
https://hal.archives-ouvertes.fr/hal-01231839https://hal.archives-ouvertes.fr
-
The thermal conductivity of silicon nitride membranes is not
sensitive to stress
Hossein Ftouni,1, 2 Christophe Blanc,1, 2 Dimitri Tainoff,1, 2 Andrew D. Fefferman,1, 2 Martial
Defoort,1, 2 Kunal J. Lulla,1, 2 Jacques Richard,1, 2 Eddy Collin,1, 2 and Olivier Bourgeois1, 2
1Institut NÉEL, CNRS, 25 avenue des Martyrs, F-38042 Grenoble, France
2Univ. Grenoble Alpes, Inst NEEL, F-38042 Grenoble, France
Abstract
We have measured the thermal properties of suspended membranes from 10 K to 300 K for two
amplitudes of internal stress (about 0.1 GPa and 1 GPa) and for two different thicknesses (50 nm
and 100 nm). The use of the original 3ω-Volklein method has allowed the extraction of both the
specific heat and the thermal conductivity of each SiN membrane over a wide temperature range.
The mechanical properties of the same substrates have been measured at helium temperatures using
nanomechanical techniques. Our measurements show that the thermal transport in freestanding
SiN membranes is not affected by the presence of internal stress. Consistently, mechanical dissipa-
tion is also unaffected even though Qs increase with increasing tensile stress. We thus demonstrate
that the theory developed by Wu and Yu [Phys. Rev. B, 84, 174109 (2011)] does not apply to
this amorphous material in this stress range. On the other hand, our results can be viewed as a
natural consequence of the ”dissipation dilution” argument [Y. L. Huang and P. R. Saulson, Rev.
Sci. Instrum. 69, 544 (1998)] which has been introduced in the context of mechanical damping.
PACS numbers: 65.60.+a,68.65.-k,62.40.+i,63.50.Lm
1
-
I. INTRODUCTION
Silicon nitride (SiN) thin films are widely used to thermally isolate sensitive thermal
detectors, for etch masking as well as layers for micro-electromechanical systems1. Indeed,
outstanding mechanical properties including very high quality factors Q2,3 can be reached
in optimized SiN material. Depending on deposition parameters, SiN films can experience
very large residual (biaxial) stress during deposition. It is thus of prime importance to
understand the role of the internal stress not only on the mechanical properties but also
on the other physical characteristics of SiN films including optical, thermal and electrical
properties. Silicon nitride has a specific place due to its amorphous nature and the study
of stress in that compound is also an issue for the fundamental understanding of its role in
the physics of glasses2.
Using the stress to tune the thermal properties of nanomaterials is one of the possi-
ble ways to design future thermal components (thermal rectifier, thermal diode, thermal
switch4 ...). This has been proposed for monocrystalline silicon5,6, as strain in silicon is
currently used to enhance electron mobility in transistors7. Since the debate on the ori-
gin of mechanical dissipation in strained glasses like SiN2,8, the question of the effect of
stress on the thermal properties, whatever its origin (internal or external), has been raised
and theoretically addressed for the case of silicon nitride9. Indeed it is well known that
stoichiometric silicon nitride (Si3N4) prepared by low pressure chemical vapor deposition
(LPCVD) contains a significant internal tensile stress (up to about 1 GPa) as compared to
regular non-stoichiometric SiN that has a very low internal stress (below 0.2 GPa). In an
attempt to explain the very high mechanical Qs, Wu and Yu9 proposed a model where the
internal losses in the material are sensitive to the stress state. Their calculations based on
this hypothesis predict that the thermal conductivity of SiN may be strongly enhanced by
the presence of stress. On the other hand, systematic mechanical measurements on high
stress SiN substrates explain the Qs through the so called ”dissipation dilution” model10:
mechanical energy is stored through the tensioning of the substrate while the dissipation is
unaffected9,11,12. However, no experiments to date compared directly similar devices made
of different SiN materials. Furthermore, thermal properties of low stress SiN have been
widely measured over a broad temperature range13–18 for different kinds of thin films and
2
-
nanomaterials, but very few experimental studies deal with the influence of stress on the
thermal transport at the nanoscale19.
In order to study the potential effect of internal stress on the thermal properties of silicon
nitride, the thermal conductivity and the specific heat have been measured as a function of
temperature for high stress (HS) and low stress (LS) SiN membranes having a thickness of
50 nm and 100 nm. These measurements are performed using the 3ω-Völklein method20–22
as described in previous papers. This technique allows the measurement of both thermal
conductivity and specific heat of a given membrane within the same experiment over a
broad temperature range. The mechanical dissipation and stress amplitude of both HS and
LS substrates are also measured at cryogenic temperatures by means of nanomechanical
resonators23. We show experimentally that thermal conduction is essentially independent
of the stress stored in this material. This is inconsistent with the hypothesis underlying
the model of Ref.9 for SiN, and corroborates the ”dissipation dilution” explanation for high
mechanical Qs.
II. SAMPLES AND EXPERIMENTAL METHODS
The thermal properties of two types of SiN membranes have been measured: high stress
stoichiometric Si3N4 and low stress SiN deposited by LPCVD. The amorphous stoichiometric
high stress (HS) Si3N4 as well as low stress (LS) SiN were grown on both sides of a silicon
substrate. The membranes were then patterned on the rear side by laser photolithography.
After removing the silicon nitride by SF6 Reactive Ion Etching, the silicon substrate on the
rear side was etched in KOH, as described in Fig. 1. The final result is a rectangular SiN
membrane obtained on the front side.
Before the thermal study, a mechanical measurement was performed to quantify the
stress present after releasing the membranes. A suspended silicon nitride beam with 100 nm
thickness, 250 nm width and 15 µm length fabricated using e-beam lithography from the
same substrate was placed in a magnetic field (see Fig. 1 for fabrication details). A sinusoidal
driving current within a 30 nm thin deposited Al layer is used to generate the Lorentz force
causing the beam’s out-of-plane oscillation24. This measurement is performed in a vacuum
of about 10−6 mbar at helium temperatures. The magnetic flux cut by the beam oscillation
3
-
FIG. 1. (colour online) Micro and nanofabrication processes of both suspended structures studied
in this work. a) fabrication process for the membrane 1), 2) The patterns of the membranes are
created by photolithography. The non protected SiN is removed by SF6 RIE etching. 3) The silicon
is anisotropically etched in a KOH solution. 4) The thermometers are obtained by a lift-off process;
the area is patterned by photolithography. 5) NbN (70 nm) is deposited by reactive sputtering.
6) The resist and NbN layer is removed using a wet procedure. b) fabrication process for the
nanowires 1) The patterns of the nanowires are created by e-beam lithography. 2) evaporation of
Al layer (30 nm) 3), 4) The non-protected SiN is removed by SF6 RIE. 5) The silicon is isotropically
etched by gazeous XeF2 ecthing.
generates a voltage which is measured using a lock-in amplifier3,25. Typical resonance curves
for the first flexure and their respective fits are shown in Fig. 2.
The expression for the nth mode resonance frequency of a stressed doubly-clamped beam
is given by8:
fn =n
2
√σ
ρh2(1)
Eq. 1 is used to calculate the stress σ within the beam, with n the mode number, ρ the
4
-
FIG. 2. (colour online) Measurement of the first flexure resonance of two suspended SiN beams
with different in-built stresses, the red points are the in-phase signal and the black points the
out-of-phase signal (both 15 µm long, 250 nm wide, 100 nm thick, see SEM picture in inset).
(a) low-stress and (b) high-stress resonance lines obtained in the linear regime. The lines are
Lorentzian fits, with full width at half height of 650 ± 50 Hz (HS, Q ≈ 25 000) and 500 ± 50 Hz
(LS, Q ≈ 14 000). We extract from the resonance frequencies the stress values of 0.85± 0.08 GPa
(HS) and 0.12± 0.05 GPa (LS). Data taken at 4.2 K in vacuum in a 840 mT magnetic field.
silicon nitride density (3 g/cm3) and h the beam length23. We find a stress value of about
0.85 GPa for the HS silicon nitride, which confirms that the membrane is still stressed after
releasing and 0.12 GPa for the LS silicon nitride. These values agree fairly well with the
manufacturer data, as they should (supplied by LIONIX). Note that both values fall in the
high-stress limit of beam theory, validating the use of Eq. (1).
A careful characterization of the setup and of the devices has been performed in order to
guarantee quantitative analysis25,26. In particular, the loading from the environment onto
the measured resonance has been characterized as a function of magnetic field24: the raw
resonance lines displayed on Fig. 2 are only about 20 % broader than the genuine intrinsic
mechanical resonances. Many devices varying shapes and stress have been measured26,
studying flexural modes up to n = 9.
The important and still unsolved issue of the mechanical dissipation shall be discussed
5
-
elsewhere36. But let us nonetheless position our findings within the state-of-the-art.
The mechanical Q factors obtained for our high-stress samples are consistent with the
litterature8,11,27–29, as are the low-stress results8,30. Obviously, these comparisons have to be
taken with care, since temperature and metallic coatings are known to influence mechanical
dissipation31–33, but the dispersion among silicon nitride wafer providers seems to be greater
or of the same order than these effects. Comparing very different types of devices had lead to
the proposition that stress could strongly influence mechanical dissipation2. Recent results
from Refs.11,12,27 on HS devices contradict this idea, and favor the ”dissipation dilution”
model first introduced in Ref.10: the mechanical Q factor increases only because the stored
(tensioning) energy increases. Indeed, tuning the stress by bending the sample a linear rela-
tionship between Q and f0 is found in Ref.8. In this respect, Fig. 2 is a natural consequence
of the ”dissipation dilution” idea: while the mechanical Q = f0/∆f increases with stress,
the linewidh ∆f (measuring mechanical dissipation) of the resonances of two geometrically
identical devices is almost unaffected. The related question we thus want to address in this
paper is how stress affects the thermal properties of silicon nitride structures.
Thermal experiments are conducted on the very same materials. As mentioned above, we
have chosen in this study the most appropriate method to measure the thermal conductivity
of large aspect ratio suspended membranes: the 3ω-Völklein method20–22. The principle of
the method consists in creating a sinusoidal Joule heating generated by an AC electrical
current at frequency ω across a transducer centered along the long axis of a rectangular
membrane. The center of the membrane is thermally isolated from the frame and hence its
temperature is free to increase.
The temperature oscillation (≈100 mK) of the membrane is at 2ω and is directly related
to its thermal properties by the amplitude and the frequency dependence of the aforemen-
tioned temperature oscillation. Since the resistance of the thermometer can be considered
as linearly dependent on temperature over that small temperature oscillation, the voltage
V = R[T (2ω)]× I(ω) will have an ohmic component at ω and a thermal component at 3ω.
By measuring the V3ω voltage appearing across the transducer as function of the frequency,
both the thermal conductivity and the specific heat of the membrane22 are inferred. The
membranes measured in this study are 300 µm wide and 1.5 mm long.
The transducer of 5 µm width and 1.5 mm length is made out of NbN whose resistance
6
-
FIG. 3. (colour online) Experimental set-up based on the Wheatstone bridge configuration, the
yellow membrane sample is on the bottom right and the reference thermometer is on the left; the
blue area of the sketch corresponds to the temperature regulated part of the Wheatstone bridge.
is strongly temperature dependent. It serves as a thermometer and a heater at the same
time37–39. For the present work, the thermometer has been designed for the 10 K to 320 K
temperature range. Typically, the resistance of the thermometer is about 100 kOhm at
room temperature with a temperature coefficient of resistance (TCR) α = dRRdT
of 10−2 K−1
at 300 K and of 0.1 K−1 at 4 K.
Since the 1ω voltage is 3 to 4 orders of magnitude higher than the 3ω voltage, a specific
Wheatstone bridge is used to reduce the 1ω component and perform thermal measurements
(see Fig. 3). The bridge consists of the measured sample with a resistance Re, which is the
NbN thermometer on the SiN membrane, the reference thermometer Rref deposited on the
bulk region of the chip which has the same geometry and deposited in the same run as the
transducer on the membrane, an adjustable resistor Rv, and an equivalent nonadjustable
resistance R1 =50 kOhm.
The general expression of the measured 3ω output Wheatstone bridge voltage can be
given by20,21:
7
-
|V rms3ω (ω)| =V rmsac αReR1 |∆T2ω|
2(Re +R1)2(2)
with α the TCR, Vac the 1ω input Wheatstone bridge voltage and |∆T2ω| the amplitude
of the temperature oscillation at 2ω of the membrane due to the sinusoidal nature of heating.
By solving the partial differential equation of the heat flux across the membrane, eq. 3
gives the relation between the thermal properties, the dimensions of the membrane and
V3ω20,21:
|V rms3ω (ω)| =α(V rmsac )
3R1R2e
4Kp (Re +R1)4 [1 + ω2 (4τ 2 + 2`4
3D2+ 4τ`
2
3D
)]1/2 (3)with Kp =
kS`
the thermal conductance and C = cS` the heat capacity of the measured
membrane, τ = CKp
the thermalization time of the membrane to the heat bath, D = kρc
the thermal diffusivity, ` half the width of the membrane and S the section of the mem-
brane (perpendicular to the heat flow). By measuring the V3ω voltage as a function of
frequency both k (in-plane thermal conductivity) and c (specific heat) of the membrane can
be extracted22.
FIG. 4. (colour online) Thermal conductivity measurement of 50 nm and 100 nm thick membranes
for both SiN low stress and high stress. The 100 nm curves of low stress and high stress show
nearly no difference.
8
-
III. EXPERIMENTAL RESULTS
The thermal conductivity of the four different membranes (50 and 100 nm, low stress and
high stress) has been measured versus temperature from 10 K to 300 K. The experimental
data of thermal conductivity are presented in the Fig. 4. As expected for amorphous mate-
rials the thermal conductivity of all membranes is continuously increasing with temperature
as observed by Queen and Hellman17. The general trend of the temperature variation of
thermal conductivity of all different SiN membranes (LS and HS) are very similar. Only the
50 nm LS membrane seems to have a slightly lower thermal conductivity at room tempera-
ture with a value approaching 2.5 W.m−1.K−1, instead of 3 W.m−1.K−1 for the others.
In all cases, values of the thermal conductivity at room temperature are approximately
3 W.m−1.K−1. These values are in accordance with most of the in plane values of thermal
conductivity measured which are displayed in table I. Indeed Jain and Goodson40 have
measured the in-plane thermal conductivity of 1.5 µm thick SiN specimens to be about
5 W.m−1.K−1. At the nanoscale, Sultan et al.41 reported thermal conductivity of 500 nm thin
films as 3-4 W.m−1.K−1 for a temperature range of 77-325 K. For 180-220 nm thick LS nitride,
Zink and Hellman16 also observed temperature variation of thermal conductivity ranging
from 0.07 to 4 W.m−1.K−1 from 3 to 300 K. The cross-plane thermal conductivity measured
by Lee and Cahill42 for less than 100 nm thickness was in the range of 0.4-0.7 W.m−1.K−1
showing severely reduced thermal conductivity, which was ascribed to the interfacial thermal
resistance. Zhang and Grigoropuolos43 also observed anomalous thickness dependence and
suggested that micro structural defects may strongly influence thermal conductivity. It is
important to note that none of the above studies measure thermal conductivity as a function
of the internal stress.
In order to verify the coherence of our experimental results, we have extracted the specific
heat from the variation of the 3ω signal versus the frequency. Generally the specific heat is
not expected to vary strongly as a function of stress at room temperature9, and consequently
it is a good test for the experiment. The results for the four different membranes are shown in
Fig. 5. The temperature variation of the specific heat is very similar for the four samples. For
both 50 and 100 nm thick membranes we observe that the specific heat tends to be slightly
higher for the case of low stress sample. But here again, the differences are insignificant
9
-
Reference Deposition Stoichiometry Stress k at 300 K c at 300 K Sample
44 LPCVD Si0.66N0.34 not measured 3.2 0.7 free st. in- plane
43 LPCVD Si1N1.1 not measured 8-10 not measured free st. out of plane
40 LPCVD Si rich low stress 4.5 0.5 free st. in-plane
41 LPCVD Si rich low stress 3.5 not measured free st. in-plane
42 PECVD-APCVD Si1N1.1 not measured 0.3 not measured out of plane
29 LPCVD not measured high stress 3.2 not measured free st. in-plane
19 LPCVD Si1N1.1 from 0 to 2.4% 2.7 (LS) to 0.4 (HS) not measured free st. in-plane
this work LS LPCVD Si1N1.1 0.2 GPa 2.5 0.8 free st. in-plane
this work HS LPCVD Si3N4 0.85 GPa 3 0.8 free st. in-plane
TABLE I. Measured values of thermal conductivity (k in W.m−1.K−1) and specific heat (c in
J/g.K) of silicon nitride having different stoichiometry and/or different stress. Our results are in
accordance with most of the studies. LS is for Low Stress, HS for High Stress and free st. for free
standing membranes.
and the specific heat is very similar for all the thicknesses and stress (low and high). The
Debye temperatures deduced from the heat capacity measurements vary from 620 to 650 K
depending on the sample which is a little lower than the commonly accepted value16. Our
measurements of thermal conductivity and specific heat demonstrate that no significant
differences occur for the thermal transport in high and low stress SiN material because even
with a stress close to 1 GPa, no modification of the phonon thermal conductivity can be
observed.
IV. DISCUSSION
High stress silicon nitride mechanical devices exhibit remarkable Q factors: inverse quality
factors Q−1 are two to three orders of magnitude lower than those of amorphous SiO2 from
4 K up to room temperature2. The true origin of mechanical dissipation in stressed SiN is
still unknown but could have connections with the thermal properties2,11,12. Even though
amorphous solids are by nature diverse in composition, these materials are characterized
by a universal behaviour of the thermal conductivity and mechanical dissipation at low
temperature (between 0.1 and 10 K)45,46. This universal behavior was initially reported
by Zeller and Pohl46 and described in terms of a phenomenological model which takes into
10
-
FIG. 5. (colour online) Specific heat measurement of 50 nm and 100 nm thick of both LS and HS
silicon nitride from low temperature (10 K) to room temperature.
account the contribution from defects referred to as two-level systems (TLS)47,48. The model
does reproduce the data, but the universality appears as a surprising coincidence which
continues to puzzle physicists2,49.
FIG. 6. (colour online) Mean free path Λ of measured samples calculated using experimental data
of specific heat and thermal conductivity. The dashed line shows the estimation of the mean free
path using the Debye specific heat.
11
-
In the theoretical work by Wu and Yu9, the starting point is to consider that the stress
(bond constraints, impurities, local defaults or even external strain) can modify either the
TLS barrier height V or the coupling between TLS and phonons denoted by γ. In this model,
it is predicted that the modification of V and γ (by taking into account the amplitude of
the stress in stoichiometric SiN) will have a significant effect on the thermal conductivity
and mechanical dissipation. Let us discuss the two cases separately. First, when the barrier
height is affected, a difference between the thermal conductivity in low and high stress should
be seen as the temperature is reduced; a factor close to five at 50 K is expected between
the thermal conductivities of LS and HS. This is clearly not observed in our measurements
since the thermal conductivity of the HS and LS membranes are very similar. We can only
point out that around 50 K the thermal conductivity is slightly different between 50 nm
and 100 nm samples, a behaviour that can be attributed to a reduction of mean free path
in the thinner membrane. Secondly, in the case of the coupling between phonons and TLS
(given by the parameter γ), an effect even larger is expected with a thermal conductivity a
factor of ten higher in the HS SiN as compared to the LS at room temperature. This could
be indeed very interesting for practical applications. Even though the stoichiometry is not
strictly identical between the low and high stress membranes, we do not observe such a big
difference in thermal conductivity. This has to be compared to the mechanical measurement
performed at 4 K, which also did not present any large differences in mechanical damping
between HS and LS devices.
We thus demonstrated negligible effect of stress on the thermal conductivity and me-
chanical dissipation in amorphous SiN. We conclude that the hypothesis of TLS in which
barrier height V or coupling constant γ is affected by stress does not apply to these mate-
rials in the present stress range. We also underline that the values of thermal conductivity
we have measured for both high stress Si3N4 and low stress SiN membranes are in perfect
accordance with most of the values already published (see table I). In order to highlight the
low temperature particularities of the phonon conductivity in these thin membranes, it is
particularly important to discuss the temperature variation of the mean free path18. Fig. 6
shows the phonon mean free path in the membranes determined from the kinetic equation
Λ = 3k/Cvs, vs being the Debye speed of sound. It has been shown in the past that this
equation can be used even at room temperature for amorphous materials by Pohl and co-
12
-
workers45. At 300 K all curves (with the exception of 50 nm LS) approach the same limit
which is two times higher than the inter-atomic spacing (0.25 nm for amorphous SiN). This
is in very good agreement with previous thermal analysis18. As the temperature decreases,
the mean free path increases rapidly to reach the order of ten nanometers at 20 K. As it can
be seen in Fig. 6, it is reasonable to ascribe the difference of thermal transport below 200 K
to a reduced mean free path in the thinner membranes.
V. CONCLUSIONS
The thermal conductivity has been measured on silicon nitride membranes having low
and high stress. The objective was to search for any effect of internal stress on the phonon
thermal conductivity and mechanical dissipation. Even though very high stress (of the order
of 1 GPa) has been evidenced in suspended stoichiometric SiN membranes by nanomechan-
ical measurements, it has been shown using very sensitive 3ω technique that the thermal
conductivity was not affected. Besides, mechanical dissipation is almost independent of
stress, even though high Qs are obtained in HS structures in accordance with the ”dissipa-
tion dilution” model. This rules out a scenario of strong increase of thermal conductivity
(and concomittantly a strond decrease of mechanical dissipation) with the presence of stress
proposed recently by Wu and Yu9, either through the increase of the barrier height of two
level systems or through the decrease of the coupling between TLS and phonons. We also
show that the thermal properties of the most commonly used silicon nitride materials are
equivalent. We then express doubts about the possible use of stress in thermal engineering
in amorphous materials.
VI. ACKNOWLEDGMENTS
We acknowledge technical supports from Nanofab, the Cryogenic and the Electronic
facilities and the Pole Capteur Thermométrique et Calorimétrie of Institut Néel for these
experiments. Funding for this project was provided by a grant from La Région Rhône-
Alpes (Cible and CMIRA), by the Agence Nationale de la Recherche (ANR) through the
project QNM no. 0404 01, by the European projects: MicroKelvin FP7 low temperature
13
-
infrastructure Grant no. 228464 and MERGING Grant no. 309150.
1 S.V. Garimella, Advances in mesoscale thermal management technologies for microelectronics.
Microelectronics Journal, 37 1165 (2006).
2 D.R. Southworth, R.A. Barton, S.S. Verbridge, B. Ilic, A.D. Fefferman, H.G. Craighead, and
J.M. Parpia, Phys. Rev. Lett. 102, 225503 (2009).
3 M. Defoort, K.J. Lulla, C. Blanc, O. Bourgeois, E. Collin, and A.D. Armour, Appl. Phys. Lett.
103, 013104 (2013).
4 N. Li, J. Ren, L. Wang, G. Zhang, P. Hanggi, B. Li, Rev. Mod. Phys. 84, 1045 (2009).
5 X. Li, K. Maute, M.L. Dunn, and R. Yang, Phys. Rev. B 81, 245318 (2010).
6 A. Paul and G. Klimeck, Appl. Phys. Lett. 99, 083115 (2011).
7 Y.J. Mii, Y.H. Xie, E.A. Fitzgerald, D. Monroe, F.A. Thiel, B.E. Weir and L.C. Feldman, Appl.
Phys. Lett. 59, 1611 (1991)
8 Scott S. Verbridge, Daniel Finkelstein Shapiro, Harold G. Craighead, and Jeevak M. Parpia,
Nano Letters 7, 1728 (2007).
9 J. Wu and C.C. Yu, Phys. Rev. B 84, 174109 (2011).
10 Y. L. Huang and P. R. Saulson, Rev. Sci. Instrum. 69, 544 (1998).
11 Q.P. Unterreithmeier, T. Faust, and J.P. Kotthaus, Phys. Rev. Lett. 105, 027205 (2010).
12 P.-L. Yu, T.P. Purdy, and C.A. Regal, Phys. Rev. Lett. 108, 083603 (2012).
13 M.M. Leivo and J.P. Pekola, Appl. Phys. Lett. 72, 1305 (1998).
14 W. Holmes, J.M. Gildemeister, and P.L. Richards, V. Kotsuboc, Appl. Phys. Lett. 72, 2250
(1998).
15 K. Schwab, E.A. Henriksen, J.M. Worlock, M.L. Roukes, Nature 404, 974 (2000).
16 B.L. Zink and F. Hellman, Solid State Commun. 129, 199 (2004).
17 D.R. Queen and F. Hellman, Rev. Sci. Instrum. 80, 063901 (2009).
18 R. Sultan, A.D. Avery, J.M. Underwood, S.J. Mason, D. Bassett, and B.L. Zink, Phys. Rev. B
87, 214305 (2013).
14
-
19 M.T. Alam, M.P. Manoharan, M.A. Haque, C. Muratore and A. Voevodin, J. Micromech.
Microeng. 22, 045001 (2012).
20 A. Sikora, H. Ftouni, J. Richard, C. Hébert, D. Eon, F. Omnès, and O. Bourgeois, Rev. Sci.
Instrum. 83, 054902 (2012).
21 A. Sikora, H. Ftouni, J. Richard, C. Hébert, D. Eon, F. Omnès, and O. Bourgeois, Rev. Sci.
Instrum. 84, 029901 (2013).
22 H. Ftouni, D. Tainoff, J. Richard, K. Lulla, J. Guidi, E. Collin, and O. Bourgeois, Rev. Sci.
Instrum. 84, 094902 (2013).
23 M. Defoort, K.J. Lulla, C. Blanc, H. Ftouni, O. Bourgeois, E. Collin, J. Low Temp. Phys. 171,
731 (2013).
24 A.N. Cleland and M.L. Roukes, Sensors and Actuators A 72, 256 (1999).
25 E. Collin, M. Defoort, K. Lulla, T. Moutonet, J.-S. Heron, O. Bourgeois, Yu. M. Bunkov, and
H. Godfrin, Rev. Sci. Instrum. 83, 045005 (2012).
26 M. Defoort, Non-linear dynamics in nano-electromechanical systems at low temperatures, PhD
thesis Université Joseph Fourier Grenoble, 2014 (unpublished). Typical parameters for our Al-
metallized HS mechanical SiN beams are (in first flexure, vacuum at 4.2 K): f0 = 0.7 MHz,
Q0 = 600 000 for 300 µm, f0 = 5 MHz, Q0 = 120 000 for 50 µm, and f0 = 16 MHz, Q0 = 25 000
for 15 µm, in good agreement with the literature34,35.
27 A. Suhel, B. D. Hauer, T. S. Biswas, K. S. D. Beach, and J. P. Davis, Appl. Phys. Lett. 100,
173111 (2012)
28 King Y. Fong, Wolfram H. P. Pernice, and Hong X. Tang, Phys. Rev. B 85, 161410(R) (2012).
29 Scott S. Verbridge, Jeevak M. Parpia, Robert B. Reichenbach, Leon M. Bellan and H. G.
Craighead, J. of Appl. Phys. 99, 124304 (2006).
30 S. Schmid, K. D. Jensen, K. H. Nielsen, and A. Boisen, Phys. Rev. B 84, 165307 (2011).
31 K.J. Lulla, M. Defoort, C. Blanc, O. Bourgeois, and E. Collin, Phys. Rev. Lett. 110, 177206
(2013).
32 E. Collin, J. Kofler, S. Lakhloufi, S. Pairis, Yu. M. Bunkov, and H. Godfrin, J. of Appl. Phys.
107, 114905 (2010).
15
-
33 A. Olkhovets, S. Evoy, D. W. Carr, J. M. Parpia, and H. G. Craighead, J. Vac. Sci. Technol. B
18(6), 3549 (2000).
34 A. Venkatesan, K. J. Lulla, M. J. Patton, A. D. Armour, C. J. Mellor, and J. R. Owers-Bradley,
Phys. Rev. B 81, 073410 (2010).
35 T.S. Biswas, A. Suhel, B.D. Hauer, A. Palomino, K.S.D. Beach, and J.P. Davis, Appl. Phys.
Lett. 101, 093105 (2012).
36 Article under preparation; see Ref.26, Sec. II.5.
37 O. Bourgeois, E. André, C. Macovei, and J. Chaussy, Rev. Sci. Instrum. 77, 126108 (2006).
38 J.S. Heron, T. Fournier, N. Mingo, and O. Bourgeois, Nano Lett. 9, 1861 (2009).
39 A.F. Lopeandia, E. André, J.-L. Garden, D. Givord, and O. Bourgeois, Rev. Sci. Instrum. 81,
053901 (2010).
40 A. Jain and K.E. Goodson, Journal of Heat Transfer, 130, 102402 (2008).
41 R. Sultan, A.D. Avery, G. Stiehl, and B.L. Zink, J. Appl. Phys. 105, 043501 (2009).
42 S.M. Lee, and D.G. Cahill, J. Appl. Phys. 81, 2590 (1997).
43 X. Zhang and C.P. Grigoropoulos, Rev. Sci. Instrum. 66, 1115 (1995).
44 C.H. Mastrangelo, Y.C. Tai and R.S. Muller, Sensors and actuators A21-A23, 856 (1990).
45 R.O. Pohl, X. Liu, E. Thompson, Rev. Mod. Phys. 74, 991 (2002).
46 R.C. Zeller, and R.O. Pohl, Phys. Rev. B 4, 2029 (1971).
47 W.A. Phillips, J. Low Temp. Phys. 7, 351 (1972).
48 P.W. Anderson, B.I. Halperin, and C.M. Varma, The Phil. Mag. 25, 1 (1972).
49 Dervis C. Vural, Anthony J. Leggett, Journal of Non-Crystalline Solids 357, 3528 (2011).
16
The thermal conductivity of silicon nitride membranes is not sensitive to stressAbstractIntroductionSamples and experimental methodsExperimental resultsDiscussionConclusionsAcknowledgmentsReferences